When the guest closes the audio device we must start dropping input
samples from JACK and zeroing the output buffer samples. Failure to do
so causes sound artifacts during operations such as guest OS reboot, and
causes a hang of the input pipeline breaking it until QEMU is restated.
Closing and reconnecting to JACK was tested during these enable/disable
calls which works well for Linux guests, however Windows re-opens the
audio hardware repeatedly even when doing simple tasks like playing a
system sounds. As such it was decided it is better to feed silence to
JACK while the device is disabled.
Signed-off-by: Geoffrey McRae <geoff@hostfission.com>
Message-id: 20200613040518.38172-6-geoff@hostfission.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This fixes a hang when there is a communications issue with the JACK
server. Simply closing the connection is enough to completely clean up
and as such we do not need to remove the ports first. As JACK uses a
socket based protocol that relies on the `select` call, if there is a
communication breakdown with the server the client library waits
forever for a response to the unregister request.
Signed-off-by: Geoffrey McRae <geoff@hostfission.com>
Message-id: 20200613040518.38172-5-geoff@hostfission.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Initial code for JACK did not support audio input and as such this
boolean was set to let QEMU know, however JACK ended up including input
support making this invalid. Further investigation shows it was invalid
to set it in the first instance anyway due to a failure on my part
understand properly what this was for when the audodev was initially
developed.
Signed-off-by: Geoffrey McRae <geoff@hostfission.com>
Message-id: 20200613040518.38172-4-geoff@hostfission.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
JACK does not provide us with the configured buffer size until after
activiation which was overriding this minimum value. JACK itself doesn't
have this minimum limitation, but the QEMU virtual hardware and as such
it must be enforced, failure to do so results in audio discontinuities.
Signed-off-by: Geoffrey McRae <geoff@hostfission.com>
Message-id: 20200613040518.38172-2-geoff@hostfission.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The buffer is the captured input to pass to backends.
As we should not modify it, mark the argument const.
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20200505132603.8575-3-f4bug@amsat.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The samples are the input to convert to u64. As we should
not modify them, mark the argument const.
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20200505132603.8575-2-f4bug@amsat.org>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Commit 571a8c522e caused the HMP wavcapture command to segfault when
processing audio data in audio_pcm_sw_write(), where a NULL
sw->hw->pcm_ops is dereferenced. This fix checks that the pointer is
valid before dereferincing it. A similar fix is also made in the
parallel function audio_pcm_sw_read().
Fixes: 571a8c522e (audio: split ctl_* functions into enable_* and
volume_*)
Signed-off-by: Bruce Rogers <brogers@suse.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-id: 20200521172931.121903-1-brogers@suse.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The code in CONV_NATURAL_FLOAT() and CLIP_NATURAL_FLOAT()
seems to use the constant 2^31-0.5 to convert float to integer
and back. But the float type lacks the required precision and
the constant used for the conversion is 2^31. This is equiva-
lent to a [-1.f, 1.f] <-> [INT32_MIN, INT32_MAX + 1] mapping.
This patch explicitly writes down the used constant. The
compiler generated code doesn't change.
The constant 2^31 has an exact float representation and the
clang 10 compiler stops complaining about an implicit int to
float conversion with a changed value.
A few notes:
- The conversion of 1.f to INT32_MAX + 1 doesn't overflow. The
type of the destination variable is int64_t.
- At a later stage one of the clip_* functions in
audio/mixeng_template.h limits INT32_MAX + 1 to the integer
range.
- The clip_natural_float_* functions in audio/mixeng.c convert
INT32_MAX and INT32_MAX + 1 to 1.f.
Buglink: https://bugs.launchpad.net/bugs/1878627
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-id: 20200523201712.23908-1-vr_qemu@t-online.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This commit adds a new audiodev backend to allow QEMU to use JACK as
both an audio sink and source.
Signed-off-by: Geoffrey McRae <geoff@hostfission.com>
Message-Id: <20200512101603.E3DB73A038E@moya.office.hostfission.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This patch prevents an underflow of variable samples in function
audio_pcm_hw_run_in(). See commit 599eac4e5a "audio:
audio_generic_get_buffer_in should honor *size". This time the
while loop in audio_pcm_hw_run_in() will terminate nevertheless,
because it seems the recording stream in Windows is always rate
limited.
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-id: 20200405075017.9901-3-vr_qemu@t-online.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
IDirectSoundCaptureBuffer_Lock() fails on Windows when called
with len = 0. Return early from dsound_get_buffer_in() in this
case.
To reproduce the warning start a linux guest. In the guest
start Audacity and you will see a lot of "Could not lock
capture buffer" warnings.
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-id: 20200405075017.9901-2-vr_qemu@t-online.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Currently the DirectSound backend fails to stop audio playback
in dsound_enable_out(). To detect a lost buffer condition
dsound_get_status_out() incorrectly uses the error code
DSERR_BUFFERLOST instead of flag DSBSTATUS_BUFFERLOST as a mask
and returns with an error. As a result dsound_enable_out()
returns early and doesn't stop playback.
To reproduce the bug start qemu on a Windows host with
-soundhw pcspk -audiodev dsound,id=audio0. On the guest
FreeDOS 1.2 command line enter beep. The image Day 1 - F-Bird
from the QEMU Advent Calendar 2018 shows the bug as well.
Buglink: https://bugs.launchpad.net/qemu/+bug/1699628
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-id: 20200405075017.9901-1-vr_qemu@t-online.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The current positive limit for the saturation nonlinearity is
only correct if the type of the result has 8 bits or less.
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-id: 20200308193321.20668-5-vr_qemu@t-online.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Currently the internal float range of the mixing engine is
[-.5f, .5f]. PulseAudio, SDL2 and libasound use a [-1.f, 1.f]
range. This means with float samples the audio playback volume
is 6dB too low and audio recording signals will be clipped in
most cases.
To avoid another scaling factor in the conv_natural_float_* and
clip_natural_float_* functions with FLOAT_MIXENG defined this
patch changes the mixing engine float range to [-1.f, 1.f].
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-id: 20200308193321.20668-4-vr_qemu@t-online.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Change the clip_natural_float_from_mono() function in
audio/mixeng.c to be consistent with the clip_*_from_mono()
functions in audio/mixeng_template.h.
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-id: 20200308193321.20668-3-vr_qemu@t-online.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This patch changes the naming scheme of the FLOAT_CONV_TO and
FLOAT_CONV_FROM macros to the scheme used in mixeng_template.h.
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-id: 20200308193321.20668-2-vr_qemu@t-online.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Fixes: 286a5d201e
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20200218094402.26625-3-philmd@redhat.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
This adds proper support for float samples in mixeng by adding a new
audio format for it.
Limitations: only native endianness is supported. None of the virtual
sound cards support float samples (it looks like most of them only
support 8 and 16 bit, only hda supports 32 bit), it is only used for the
audio backends (i.e. host side).
Signed-off-by: Kővágó, Zoltán <DirtY.iCE.hu@gmail.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Message-id: 8a8b0b5698401b78d3c4c8ec90aef83b95babb06.1580672076.git.DirtY.iCE.hu@gmail.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
There are reports that since commit 2ceb8240fa "coreaudio: port
to the new audio backend api" audio playback with CoreAudio is
broken. This patch reverts some parts the commit.
Because of changes in the audio subsystem the audio clip
function in v4.1.0 of coreaudio.c had to be moved to mixeng.c
and the generic buffer management code needed a hint about the
size of the float type.
This patch is based on a patch from Zoltán Kővágó found at
https://lists.nongnu.org/archive/html/qemu-devel/2020-01/msg02142.html.
Fixes: 2ceb8240fa "coreaudio: port to the new audio backend api"
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-id: 20200202140641.4737-1-vr_qemu@t-online.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Windows (unlike wine) bails out when IDirectSoundBuffer8::Lock is called
with zero length. Also, hw->pos_emul handling was incorrect when
calling this function for the first time.
Signed-off-by: Kővágó, Zoltán <DirtY.iCE.hu@gmail.com>
Reported-by: KJ Liew <liewkj@yahoo.com>
Tested-by: Howard Spoelstra <hsp.cat7@gmail.com>
Message-id: fe9744216d9d421a2dbb09bcf5fa0dbd18f77ac5.1580684275.git.DirtY.iCE.hu@gmail.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The function generic_get_buffer_in currently ignores the *size
parameter and may return a buffer larger than *size.
As a result the variable samples in function
audio_pcm_hw_run_in may underflow. The while loop then most
likely will never termiate.
Buglink: http://bugs.debian.org/948658
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20200123074943.6699-9-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Currently there is no way to disable poll mode in
oss_enable_out and oss_enable_in when it was enabled before.
The enable code path always resets the poll mode state variable.
Fixes: b027a538c6 "oss: Remove unused error handling of qemu_set_fd_handler"
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20200123074943.6699-8-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
With audiodev parameter out.mixing-engine=off hw->mix_buf is
NULL. This patch reverts a small part of dc88e38fa7 "audio:
unify input and output mixeng buffer management".
To reproduce the problem start qemu with
-audiodev oss,id=audio0,try-mmap=on,out.mixing-engine=off
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20200123074943.6699-6-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The combined generic buffer management code and buffer run out
code in function audio_generic_put_buffer_out has a problematic
behaviour. A few hundred milliseconds after playback starts the
mixing buffer and the generic buffer are nearly full and the
following pattern can be seen.
On first call of audio_pcm_hw_run_out the buffer run code in
audio_generic_put_buffer_out writes some data to the audio
hardware but the generic buffer will fill faster and is full
when audio_pcm_hw_run_out returns. This is because emulated
audio devices can produce playback data at a higher rate than
the audio backend hardware consumes this data.
On next call of audio_pcm_hw_run_out the buffer run code in
audio_generic_put_buffer_out writes some data to the audio
hardware but no audio data is transferred to the generic buffer
because the buffer is already full.
Then the pattern repeats. For the emulated audio device this
looks like the audio timer period has doubled.
This patch splits the combined generic buffer management code
and buffer run out code and calls the buffer run out code after
buffer management code to break this pattern.
The bug report is for the wav audio backend. But the problem is
not limited to this backend. All audio backends which use the
audio_generic_put_buffer_out function show this problem.
Buglink: https://bugs.launchpad.net/qemu/+bug/1858488
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20200123074943.6699-5-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
With audiodev parameter out.mixing-engine=off hw->mix_buf is
NULL. This leads to a segmentation fault in
AUD_get_buffer_size_out. This patch reverts a small part of
dc88e38fa7 "audio: unify input and output mixeng buffer
management".
To reproduce the problem start qemu with
-soundhw adlib -audiodev pa,id=audio0,out.mixing-engine=off
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20200123074943.6699-4-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The unused variables were last used before commit 49ddd7e122
"paaudio: port to the new audio backend api".
Fixes: 49ddd7e122
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20200123074943.6699-3-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
It seems the function audio_generic_read started as a copy of
function audio_generic_write and some necessary changes were
forgotten. Fix the mixed up source and destination pointers and
rename misnamed variables.
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20200123074943.6699-2-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The pcm_ops function put_buffer_out expects the returned pointer
of function get_buffer_out as argument. Fix this.
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20200123074943.6699-1-vr_qemu@t-online.de>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Fixes: 3ba4066d08 ("ossaudio: port to the new audio backend api")
Reported-by: ziming zhang <ezrakiez@gmail.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-Id: <20200120101804.29578-1-kraxel@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
When building with GCC9 using CFLAG -Wimplicit-fallthrough=2 we get:
audio/audio.c: In function ‘audio_pcm_init_info’:
audio/audio.c:306:14: error: this statement may fall through [-Werror=implicit-fallthrough=]
306 | sign = 1;
| ~~~~~^~~
audio/audio.c:307:5: note: here
307 | case AUDIO_FORMAT_U8:
| ^~~~
cc1: all warnings being treated as errors
Similarly to e46349414, add the missing fall through comment to
hint GCC.
Fixes: 2b9cce8c8c
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Aleksandar Markovic <amarkovic@wavecomp.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Message-Id: <20191218192526.13845-2-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Tell the compiler to do a 32bit * 32bit -> 64bit multiplication
because period_ticks is a 64bit variable. The overflow occurs
for audio timer periods larger than 4294967us.
Fixes: be1092afa0 "audio: fix audio timer rate conversion bug"
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-id: 8893a235-66a8-8fbe-7d95-862e29da90b1@t-online.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Don't call pa_stream_peek before the recording stream is ready.
Information to reproduce the problem.
Start and stop Audacity in the guest several times because the
problem is racy.
libvirt log file:
-audiodev pa,id=audio0,server=localhost,out.latency=30000,
out.mixing-engine=off,in.mixing-engine=off \
-sandbox on,obsolete=deny,elevateprivileges=deny,spawn=deny,
resourcecontrol=deny \
-msg timestamp=on
: Domain id=4 is tainted: custom-argv
char device redirected to /dev/pts/1 (label charserial0)
audio: Device pcspk: audiodev default parameter is deprecated,
please specify audiodev=audio0
audio: Device hda: audiodev default parameter is deprecated,
please specify audiodev=audio0
pulseaudio: pa_stream_peek failed
pulseaudio: Reason: Bad state
pulseaudio: pa_stream_peek failed
pulseaudio: Reason: Bad state
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-id: 20200104091122.13971-5-vr_qemu@t-online.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
There is no guarantee a single call to pa_stream_peek every
timer_period microseconds can read a recording stream faster
than the data gets produced at the source. Let qpa_read try to
drain the recording stream.
To reproduce the problem:
Start qemu with -audiodev pa,id=audio0,in.mixing-engine=off
On the host connect the qemu recording stream to the monitor of
a hardware output device. While the problem can also be seen
with a hardware input device, it's obvious with the monitor of
a hardware output device.
In the guest start audio recording with audacity and notice the
slow recording data rate.
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-id: 20200104091122.13971-4-vr_qemu@t-online.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Every call to pa_stream_peek which returns a data length > 0
should have a corresponding pa_stream_drop. A call to qpa_read
does not necessarily call pa_stream_drop immediately after a
call to pa_stream_peek. Test in qpa_fini_in if a last
pa_stream_drop is needed.
This prevents following messages in the libvirt log file after
a recording stream gets closed and a new one opened.
pulseaudio: pa_stream_drop failed
pulseaudio: Reason: Bad state
pulseaudio: pa_stream_drop failed
pulseaudio: Reason: Bad state
To reproduce start qemu with
-audiodev pa,id=audio0,in.mixing-engine=off
and in the guest start and stop Audacity several times.
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-id: 20200104091122.13971-3-vr_qemu@t-online.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
With current code audio recording with all audio backends
except PulseAudio and DirectSound is broken. The generic audio
recording buffer management forgot to update the current read
position after a read.
Fixes: ff095e5231 "audio: api for mixeng code free backends"
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Zoltán Kővágó <DirtY.iCE.hu@gmail.com>
Message-id: 2fc947cf-7b42-de68-3f11-cbcf1c096be9@t-online.de
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Which currently only means removing some checks. Old code won't require
more than two channels, but new code will need it.
Signed-off-by: Kővágó, Zoltán <DirtY.iCE.hu@gmail.com>
Message-id: 7e53be1f97e939ed3bb729ef39e76b775643118a.1570996490.git.DirtY.iCE.hu@gmail.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
The bit shifting trick worked because the number of bytes per frame was
always a power-of-two (since QEMU only supports mono, stereo and 8, 16
and 32 bit samples). But if we want to add support for surround sound,
this no longer holds true.
Signed-off-by: Kővágó, Zoltán <DirtY.iCE.hu@gmail.com>
Message-id: 1351fd9bcce0ff20d81850c5292722194329de02.1570996490.git.DirtY.iCE.hu@gmail.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This can be used to identify stream in tools like pavucontrol when one
creates multiple -audiodevs or runs multiple qemu instances.
Signed-off-by: Kővágó, Zoltán <DirtY.iCE.hu@gmail.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Message-id: 2d6e337c474ac84172d0809e6959c26b21d48120.1568157545.git.DirtY.iCE.hu@gmail.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Connection name was previously erroneously set to the server socket
path, while connection names were simply "qemu". After this patch, the
connection name will be the vm name (falling back to "qemu" if not
specified), while stream names will be the audiodev's id.
Signed-off-by: Kővágó, Zoltán <DirtY.iCE.hu@gmail.com>
Message-id: 3d139426031a400a68d440608ba5e43f0e116cd8.1568157545.git.DirtY.iCE.hu@gmail.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This way we no longer need vararg functions, improving compile time
error detection. Also now it's possible to check actually what commands
are supported, without needing to manually update ctl_caps.
Signed-off-by: Kővágó, Zoltán <DirtY.iCE.hu@gmail.com>
Message-id: 2b08b3773569c5be055d0a0fb2f29ff64e79f0f4.1568927990.git.DirtY.iCE.hu@gmail.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
This commit removes the ad-hoc rate-limiting code from noaudio and
wavaudio, and replaces them with a (slightly modified) code from
spiceaudio. This way multiple write calls (for example when the
circular buffer wraps around) do not cause problems.
Signed-off-by: Kővágó, Zoltán <DirtY.iCE.hu@gmail.com>
Message-id: fd0fe5b95b13fa26d09ae77a72f99d0ea411de14.1568927990.git.DirtY.iCE.hu@gmail.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Usage notes: hw->samples became hw->{mix,conv}_buf->size, except before
initialization (audio_pcm_hw_alloc_resources_*), hw->samples gives the
initial size of the STSampleBuffer. The next commit tries to fix this
inconsistency.
Signed-off-by: Kővágó, Zoltán <DirtY.iCE.hu@gmail.com>
Message-id: a78caeb2eeb6348ecb45bb2c81709570ef8ac5b3.1568927990.git.DirtY.iCE.hu@gmail.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>